AIME 2025

Homepage and repository Homepage: https://matharena.ai/ Repository: https://github.com/eth-sri/matharena Dataset Summary This dataset contains the questions from AIME 2025 used for the MathArena Leaderboard Data Fields The dataset contains the following fields: problem_idx (int64): Problem index within the correspondin

Papers using AIME 2025 (13)

Cliff Tokens: Identifying Single-Token Failure Triggers in LLM Mathematical Reasoning2026Good SFT Optimizes for SFT, Better SFT Prepares for Reinforcement Learning2026STEP3-VL-10B Technical Report2026Apriel-Reasoner: RL Post-Training for General-Purpose and Efficient Reasoning2026TERMINATOR: Learning Optimal Exit Points for Early Stopping in Chain-of-Thought Reasoning2026Reflective Confidence: Correcting Reasoning Flaws via Online Self-Correction2025BAPO: Stabilizing Off-Policy Reinforcement Learning for LLMs via Balanced Policy Optimization with Adaptive Clipping2025Beyond the Last Answer: Your Reasoning Trace Uncovers More than You Think2025Solve-Detect-Verify: Inference-Time Scaling with Flexible Generative Verifier2025Not All Correct Answers Are Equal: Why Your Distillation Source Matters2025DeepPrune: Parallel Scaling without Inter-trace Redundancy2025BAPO: Stabilizing Off-Policy Reinforcement Learning for LLMs via Balanced Policy Optimization with Adaptive Clipping2025UloRL:An Ultra-Long Output Reinforcement Learning Approach for Advancing Large Language Models' Reasoning Abilities2025